This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
Okay I'm attaching a patch with those changes, and I'll re-run all the tests. Assuming they all pass, may I commit this patch?
Caroline Tice <ctice@apple.com> writes:
I'd prefer that this block of operations be made conditional on flag_reorder_blocks_and_partition. If the flag is false, it's going to be scanning the entire insn stream several times for no benefit.
This code is inside fix_edges_for_rarely_executed_code, which is only called from partition_hot_cold_basic_blocks, which is only called if flag_reorder_blocks_and_partition is true. Therefore it seems redundant to put a check for that flag here...
Ah, ok, yes, it's redundant then.
Why not just "cfun->unlikely_text_section_name = UNLIKELY_EXECUTED_TEXT_SECTION_NAME"? You shouldn't have any name encoding here to strip. And this has the same memory leak as the then-block.
If you are sure there won't be any encoding to strip, I'll be happy to do that. And to fix the memory leak...
The various _SECTION_NAME macros are defined not to carry name encoding, as I understand it.
zw
Attachment:
gcc5-hot-cold.txt
Description: Text document
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |